ecosystem ecology - original paper · 2017. 8. 29. · ecosystem ecology - original paper...
TRANSCRIPT
ECOSYSTEM ECOLOGY - ORIGINAL PAPER
Intra-annual variability of anatomical structure and d13C valueswithin tree rings of spruce and pine in alpine, temperateand boreal Europe
Eugene A. Vaganov Æ Ernst-Detlef Schulze ÆMarina V. Skomarkova Æ Alexander Knohl ÆWilli A. Brand Æ Christiane Roscher
Received: 20 August 2007 / Accepted: 13 July 2009 / Published online: 4 August 2009
� The Author(s) 2009. This article is published with open access at Springerlink.com
Abstract Tree-ring width, wood density, anatomical
structure and 13C/12C ratios expressed as d13C-values of
whole wood of Picea abies were investigated for trees
growing in closed canopy forest stands. Samples were col-
lected from the alpine Renon site in North Italy, the lowland
Hainich site in Central Germany and the boreal Flakaliden
site in North Sweden. In addition, Pinus cembra was studied
at the alpine site and Pinus sylvestris at the boreal site. The
density profiles of tree rings were measured using the
DENDRO-2003 densitometer, d13C was measured using
high-resolution laser-ablation-combustion-gas chromatog-
raphy-infra-red mass spectrometry and anatomical charac-
teristics of tree rings (tracheid diameter, cell-wall thickness,
cell-wall area and cell-lumen area) were measured using an
image analyzer. Based on long-term statistics, climatic
variables, such as temperature, precipitation, solar radiation
and vapor pressure deficit, explained\20% of the variation
in tree-ring width and wood density over consecutive years,
while 29–58% of the variation in tree-ring width were
explained by autocorrelation between tree rings. An inten-
sive study of tree rings between 1999 and 2003 revealed that
tree ring width and d13C-values of whole wood were
significantly correlated with length of the growing season,
net radiation and vapor pressure deficit. The d13C-values
were not correlated with precipitation or temperature.
A highly significant correlation was also found between
d13C of the early wood of one year and the late wood of the
previous year, indicating a carry-over effect of the growing
conditions of the previous season on current wood produc-
tion. This latter effect may explain the high autocorrelation
of long-term tree-ring statistics. The pattern, however, was
complex, showing stepwise decreases as well as stepwise
increases in the d13C between late wood and early wood.
The results are interpreted in the context of the biochemistry
of wood formation and its linkage to storage products. It is
clear that the relations between d13C and tree-ring width and
climate are multi-factorial in seasonal climates.
Keywords Dendrochonology � Carbohydrate storage �Picea abies � Pinus cembra � Pinus sylvestris �Tracheid lumen area � Wood density
Introduction
Tree rings are generally considered to be archives of past
climatic conditions and of changes in the environment
(Schweingruber 1996; Wimmer 2002), especially when
trees were growing under extreme conditions in which
changes in climate dominated the response. Despite a long
history of climate reconstructions (Fritts 1966), there is still
some uncertainty regarding the specific effects of certain
climate parameters, especially temperature and drought, on
tree-ring width and wood anatomy (Vaganov et al. 2006).
This uncertainty is based on the knowledge that seasonal
and annual changes in tree-ring anatomy (tree-ring width,
number of cells, size of cells, cell-wall thickness and wood
Communicated by Dan Yakir.
E. A. Vaganov � M. V. Skomarkova
Institute of Forest SB RAS, Akademgorodok,
660036 Krasnoyarsk, Russia
E.-D. Schulze (&) � W. A. Brand � C. Roscher
Max-Planck Institute for Biogeochemistry,
Box 100164, 07701 Jena, Germany
e-mail: [email protected]
A. Knohl
Department of Plant Science, ETH Zurich,
Universitatsstr. 2, 8092 Zurich, Switzerland
123
Oecologia (2009) 161:729–745
DOI 10.1007/s00442-009-1421-y
density) reflect not only processes at the time of xylem-cell
production and maturation, but also additional physiologi-
cal controls and the availability of storage products at the
time of wood formation (Hemming et al. 2001). In addition,
the conditions for wood formation may be different for
solitary growing trees under extreme environmental con-
ditions, which have been generally sampled for tree ring
studies, than for trees growing in forest stands of closed
canopies where competition for water, nutrients and space
in addition to climate be important. Also, single events,
such as late frost or heavy fruiting, may affect wood growth
more than average climatic conditions in forest stands.
In addition to wood anatomy, the isotope composition of
tree rings may serve as an additional parameter to expand
the possibilities of interpreting tree-ring width. The 13C/12C
ratios of wood (cellulose) have also been shown to reflect
climatic conditions (Wilson and Grinsted 1977; Leavitt
1993; McNulty and Swank 1995; Duquesnay et al. 1998;
McCarroll and Loader 2004). The carbon isotope ratios of
carbohydrates are determined by carbon dioxide (CO2)
assimilation and stomatal conductance during the growing
season (Francey and Farquhar 1982; Farquhar et al. 1989;
Brugnoli and Farquhar 2000). However, these ratios in tree
rings may also be influenced by isotopic discrimination
during storage and re-mobilization of storage products,
including various carbohydrates, lipids and proteins, in
leaves (Gleixner et al. 1998; Gessler et al. 2008) and stems
(Skomarkova et al. 2006). Storage effects are most
important at the end and at the beginning of the growing
season (Helle and Schleser 2004; Scartazza et al. 2004;
Schulze et al. 2004), although the interactions between
storage products and growth are not fully understood.
A carry-over effect may lead to autocorrelations among
years and mask a climatic signal (Hill et al. 1995;
Monserud and Marshall 2001; Kagawa et al. 2006a, b).
The aims of the study reported here were (1) to determine
the correlation between climate parameters and tree-ring
width (TRW) and wood anatomy (wood density, cell-wall
thickness, cell-lumen area) at three sites ranging between
boreal and sub-Mediterranean alpine climates; (2) to com-
pare the long-term (decadal) correlations between tree rings
and climate parameters with short-term observations; (3) to
analyze the seasonal variation of tree-ring d13C in trees
exhibiting different growth rates; (4) to investigate relations
between present and past years on tree-ring parameters.
Materials and methods
Experimental sites and tree samples
The three study sites are (Table 1; see also http://www.
bgc-jena.mpg.de/public/carboeur/sites/index-s.html): (1) an
alpine forest in North Italy (Renon, REN), (2) a lowland
forest in Germany (Hainich, HAI) and (3) a boreal forest in
North Sweden (Flakaliden, FLA). Annual temperatures
were highest in HAI and lowest in FLA, and the length of
the growing season ([5�C) was longest (211 days) in
Germany, 20% shorter at the alpine site and 34% shorter at
the boreal site. Annual precipitation and summer rainfall
decreased from the alpine REN site to the boreal FLA site.
The investigated stands differed in age and stand density.
The distribution of forests at the global scale is related to
precipitation, but the growth of trees is not only determined
by climate but also by nutrition (Linder 1995; Schulze et al.
2006). Table 1 shows that the boreal site has a much lower
nitrogen concentration in its needles than the other sites
and that the needle nitrogen level reached at FLA indicates
deficiency (Oren and Schulze 1989).
Soil conditions differed between sites. The boreal site at
FLA is a glacial moraine with a shallow organic layer on
unsorted stony material in the whole soil horizon. The
lowland site at HAI is a brown earth developed on clay that
originated from the weathering of limestone. The alpine
site REN is a shallow organic layer that developed on
limestone boulders.
In accordance with the climate analysis of Walter and
Lieth (1967), all sites have a positive hydrological balance.
During the time interval studied in this investigation
(Table 2; 1999–2003), the seasonal climate showed very
little variation in early season temperatures, but tempera-
tures were colder during the autumn at the boreal site.
Average daytime vapor pressure deficit was highest at the
HAI site. Early season precipitation was lowest at the
boreal site, but low rainfall in the spring is generally
considered to be compensated for by snow melt in boreal
climates. Late season rainfall was lowest at HAI. Solar
radiation was highest at the alpine site REN. Due to the
long daylight in the northern summer, solar radiation was
higher at FLA than at HAI in the early growing season, but
this decreased in the late summer.
The seasonal averages do not show the inter-annual var-
iability of climate, which is important to this study (Fig. 1).
Maximum temperatures increased at all sites between 1999
and 2003. The length of the growing season (days with
average temperature[5�C) did not show large inter-annual
variability, while precipitation showed the largest variabil-
ity, with drought being more apparent later in the season.
Years with precipitation during the growing season below
the long-term average were considered to be dry years. At
FLA, 1999 and 2002 were considered to be dry years, with
rainfall up to 30% below average; at HAI and REN, 2001
and 2003 were considered to be meteorologically dry years.
Since the CarboEurope climate data cover only 5 years,
additionally long-term meteorological data (daily temper-
ature and precipitation) were used from the following
730 Oecologia (2009) 161:729–745
123
weather stations: Costalovara (46�520N, 11�420E, 1,250 m
a.s.l., data for 1989–2003), Soprabolzano (46�520N,
11�400E, 1,206 m a.s.l., data for 1931–1980), Leinefelde
(51�200N, 10�220E, 440 m a.s.l., data for the 1957–2003)
and Stensele (65�040N, 17�100E, 330 m a.s.l., data for
1918–2001).
Tree sampling and dendrochronological analysis
Wood samples were collected by coring one core per tree at
breast height (1.3 m) with a 5-mm-diameter increment
borer. At Renon, eight cores of Norway spruce [Picea
abies (L.) H. Karst] and three cores of Swiss stone pine
(Pinus cembra L.) were used to measure wood density, and
five cores of each species were used to measure d13C and
anatomy. For Hainich, three cores each of Norway spruce
were used to measure d13C and anatomy, respectively. For
Flakaliden, three cores each of Norway spruce and Scotch
pine (Pinus sylvestris L.) were used for d13C and anatomy,
respectively. The statistical data of wood density and
tree-ring width chronologies are based on International
Tree-Ring Data Bank (ITRDB; FH Schweingruber) for the
sites in Germany and Sweden.
Tree-ring width was measured on all cores using a semi-
automatic device LINTAB-III (Cook and Kairiuktis 1990;
Rinn 1996; Vaganov et al. 1996). Density profiles of tree
rings were measured following Schweingruber (1988)
using a densitometer (Dendro-2003; Walsch Electronics,
Zurich, Switzerland).
We also used the existing in ITRDB tree-ring width and
maximum density chronologies obtained by FH Schwe-
ingruber at the end of the 1970s for sites close to the study
areas. The analog sites of HAI and FLA are Andreasberg
Harz Schlucht (900 m NN) and Gallejour Glommerstark
(480 m NN), respectively (http://www.ncdc.noaa.gov/
paleo/ftp-treering.html).
Cross-dating of individual cores was carried out based
on tree-ring width (Schweingruber 1988) using the pro-
gram COFECHA (Cook and Peters 1981; Holmes 1992).
Absolute values of each tree-ring width were standardized
by calculating the ratio of the tree-ring deviation in a
certain year and the long-term average of tree-ring width to
remove the effect of age and diameter on tree-ring width
(Shiyatov 1986; Cook et al. 1990; Vaganov et al. 1996).
A spline function was used to determine the long-term
trend for each tree core using the program ARSTAN (Cook
et al. 1990). The following parameters were calculated
according to Schweingruber (1988) as:
Table 1 Climate and stand characteristics of the Renon site (Cescatti and Marcolla 2004), the Hainich site (Knohl et al. 2003) and the Flakaliden
site (Linder 1995; Lundmark et al. 1998)
Site Renon (North Italy) Hainich (Central Germany) Flakaliden (North Sweden)
Longitude 46�360 51�040 64�070
Latitude 11�280 10�270 19�270
Elevation (m) 1,730 445 310
Length of growing seasona ([5�C) (day) 30 April–14 October (167) 31 March–27 October (211) 9 May–27 September (141)
Average annual temperaturea (�C) 3.8 6.8 1.9
Precipitation, annual (mm) 1,008 780 590
Species (no. of samples) Picea abies (13) Pinus cembra (8) P. abies (5) P. abies (5) Pinus sylvestris (5)
Age (years) 182 80 31
Diameter at breast height (cm) 17 40 8
Height (m) 29 31 6
Stand density (trees/ha) 745 679 2100
Needle nitrogen concentration (mmol g-1) 0.84 0.94 0.68
a Temperature data were corrected according the differences between the elevation of the studied site (1700 m) and that of the meteorological
station (1210 m)
Table 2 Comparison of early and late season climate between sites
Parameter Renon Hainich Flakaliden
Temperature–MJJ (�C) 14.8 15.2 11.6
Temperature–ASO (�C) 12.5 14.8 7.8
Vapor pressure deficit–MJJ (hPa) 4.2 5.8 4.7
Vapor pressure deficit– ASO (hPa) 3.7 4.5 2.6
Precipitation–MJJ (mm) 324 214 209
Precipitation–ASO (mm) 242 177 202
Precipitation-growing season (mm) 566 391 411
Solar radiation-MJJ (MJ m-2) 967 621 665
Solar radiation–ASO (MJ m-2) 686 491 299
MJJ, May–June–July; ASO, August–September–October
Study interval: 1999–2003
Oecologia (2009) 161:729–745 731
123
1. The standard deviation of relative TRW and maximum
density, indicating a measure of tree-ring response to
inter-annual climate variability.
2. The ‘‘interserial correlation’’ describes the variation in
a tree-ring parameter in a certain year between all
sample trees of a stand. This parameter provides
information on the heterogeneity of the stand and
indicates the similarity of tree growth during inter-
annual changes in climate. A high number indicates a
high similarity between trees.
3. The common variance of the first eigenvector
explains the fraction of the variability in tree-ring
parameters by environmental factors common to all
trees at each site (Fritts 1976; Cook et al. 1990).
A high number indicates a strong climatic effect on
all trees of a site.
4. The first-order autocorrelation explains the correlation
between the tree-ring width in the previous year (t-1)
and ring width in the year under investigation (t).
A large number represents a strong effect of the past
year on the current year’s tree ring.
The effect of temperature and precipitation on the inter-
annual variability of tree-ring growth and wood density
was estimated through correlations between the chronolo-
gies and monthly climatic data from the above mentioned
meteorological stations. These calculations were performed
for each month of 12 months (from October of the previous
year to September of the current year) using the program
RESPONSE (Holmes 1983).
Image analysis of tracheid dimensions
Wood cross sections were cut with a microtome (20 lm)
and stained by methylene blue (Furst 1979; Vaganov et al.
1985). Tracheid dimension in tree rings was determined
using an Image analysis system (Carl Zeiss, Jena; Munro
et al. 1996; Vaganov et al. 2006). Radial cell diameter (DR),
cell-wall thickness (CWT) and tangential cell diameter
(DT) were measured in five radial series of cells from the
outside border in each tree ring to the inside border for each
year. The five radial series were 100–150 lm (three to five
cell rows) apart. Assuming a rectangular shape of the
tracheid in cross section, these data were used to calculate
cell-wall area: CWTarea ¼ 2CWT ðDT þ DR � 2CWTÞ,and cell-lumen area: LUMarea ¼ DR � DT � CWTarea.
Since tree-ring width varies between years and species,
measurements were normalized to a standard number of
cells (Vaganov 1990).
Seasonal course of anatomy and isotopes
Tracheids of conifers are organized in rows in which each
tracheid was produced by the cambial zone consecutively
after the previous one. As such, the position (from the first
early wood cell to last late wood cell) of each tracheid in a
row is related to a temporal scale. In order to fit the ana-
tomical data measured within a tree ring with calendar
dates (dates of season) for some samples, we used the
approach described in detail by Vaganov et al. (2006)
1999 2000 2001 2002 2003
Hainich
A M J J A S O0
50
100
150
200
A M J J A S O A M J J A S O A M J J A S O A M J J A S O0
5
10
15
20470 mm 343 mm 442 mm 594 mm 236 mm
Renon
A M J J A S O
Pre
cipi
tatio
n (m
m)
0
50
100
150
200
A M J J A S O A M J J A S O A M J J A S O A M J J A S O
Air
Tem
pera
ture
(°C
)
0
5
10
15
20753 mm 661 mm 563 mm 692 mm 470 mm
Flakaliden
A M J J A S O0
50
100
150
200
A M J J A S O A M J J A S O A M J J A S O A M J J A S O0
5
10
15
20356 mm 565 mm 557 mm 290 mm 422 mm
Fig. 1 Inter-annual variability in temperature and rainfall at the three study sites for the period of 1999–2003. Gray shadow indicates dry years,
horizontal bars indicate the meteorological length of the growing season ([5�C daily temperature)
732 Oecologia (2009) 161:729–745
123
which transforms the tracheid radial size to the dates of
their production by the cambium based on two main
assumptions: (1) a linear relationship between growth rate
and radial tracheid size, and (2) a known duration of the
growth season. In tree-ring studies, the length of the growth
season is the period with an average daily air temperature
above 5�C (Mikola 1962; Leikola 1969; Hughes et al.
1999; Jones and Briffa 1992; Kirdyanov et al. 2003). Based
on this assumption, the linear sequence of tracheid mea-
surements is transformed into a seasonal time scale for
each consequent tracheid in the row, accounting for a delay
of 25 days, which is needed for cells to reach a final
diameter and to start with secondary cell-wall growth
(Vaganov et al. 2006).
Carbon isotope analysis
Carbon isotope ratios were determined using a laser abla-
tion-combustion line coupled to an isotope-ratio-mass-
spectrometer (YAG 266 nm UV laser; Merchantek–New
Wave, Fremont, CA, coupled to a Finnigan Delta ?XL) as
described in detail by Schulze et al. (2004). The exact
location of each ablation spot was visualized with a camera
mounted directly on the laser ablation station. The wood
samples were enclosed in an aluminum chamber with a
quartz window, which was flushed with helium as carrier
gas. For quantitative combustion of the ablation particles to
CO2 and H2O, the sample was passed through an Al2O3
tube at 700�C containing CuO wire as the oxygen source.
The reaction gas further passed through a gas chromatog-
raphy (GC) column (HayeSep D, Bandera, TX) to separate
CO2 from other gases. Water was removed with an on-line
Nafion water trap. Carbon-stable isotope ratios were
expressed in the d notation on the VPDB scale using
NBS22 with a value of -30.03% as the scale anchor. The
spatial resolution of a laser shot is about 70 lm. A series of
four to five shots was needed at one location to collect
sufficient material for analysis. The series of shots was
repeated radially along the same line every 120 lm, which
resulted in ten to fifteen data points per tree ring. Profiles of
d13C were measured for a 19-year period.
The d13C values within a tree ring varying within and
between years were normalized by calculating the devia-
tion of the measured value at time t from the average of
that same year (d13Ct - d13Cavt).
In a separate experiment, the d13C value of ray cells
and of adjacent tracheids were measured in the last cells of
late wood and in the first cells of early wood in order to
detect changes which may be related to carbohydrate
storage.
We use the d13C notation rather than discrimination
(D) because we did not measure the atmospheric d13C.
Furthermore, the reconstruction of all discrimination steps
during wood formation was not the objective of our study
and, therefore, the comparison of d13C values seemed to be
more appropriate. The d13C-values represent the isotope
ratios of natural wood, including lignin. Schulze et al.
(2004) showed that there is an offset between cellulose and
natural wood of 1.5% in conifers, which is constant and
independent of season. Eglin et al. (2008) also confirmed
that biochemical composition influences the variability in
the carbon isotope composition of tree rings by up to a
maximum of only 5%. Consequently the investigation of
whole wood samples seems to be appropriate (Harlow et al.
2006; Eglin et al. 2008).
Statistical analysis
Relationships between wood anatomy and carbon isotope
ratios and their dependency on climatic variables were
analyzed with linear mixed-effects models using the lmer
function in the R package lme4 (Bates and Sakar 2006) to
account for unbalances in the data set due to different
numbers of investigated trees per site and study years.
A hierarchical series of models was fitted sequentially,
entering site, wood anatomical or climatic variables and
their interactions, with site as fixed effects. Tree identity
and study year were included as grouping factors in a
nested sequence in the random term. The maximum like-
lihood method was applied, and likelihood ratio statistics
were used to appraise model improvement and to test for
statistical significance of the fixed effects.
Results
A typical pattern of the anatomical parameters and carbon
isotope ratios of the 2001 tree ring of spruce from REN
(Fig. 2) shows that the radial size of cells decreases slowly
in early spring wood and faster in late early wood and late
wood. Early and late wood are clearly visible in the
density slide. At the same time, CWT and area increase,
reaching a maximum in the first cells of late wood and
decreasing again before the cessation of growth at the end
of the season. Changes in lumen area parallel the changes
in radial cell size. Wood density can be seen to increase in
early wood slower than in late wood, reaching a maximum
in the last cells of late wood. Values of d13C increase
gently in early wood from the beginning of the growth
ring to the transition zone between early and late wood
and then decrease sharply in late wood. This decrease in
d13C (more negative values indicate open stomata) in late
wood was unexpected because precipitation was low in
August.
Oecologia (2009) 161:729–745 733
123
Long-term relations between tree-ring width, maximum
density and climate
Standardized tree-ring width and maximum wood density
of spruce and pine species varied with a standard deviation
of up to 20%, and they were similar between both species
studied (Table 3; Fig. 3). Both species showed higher
inter-annual variability for tree-ring width than for maxi-
mum wood density. The standard deviation of the
standardized tree-ring width, indicating inter-annual vari-
ability, was lower at the alpine REN site [standard devia-
tion (SD) 10%] than at the boreal FLA and the lowland
HAI site (SD 20%; Table 3). Inter-annual variability of
wood density was very small (6–7%) and similar for all
sites and species.
The inter-serial correlation (Table 3) that compares
the variation between trees is similar for maximum density
and for tree-ring width. The standard deviation of the
0
100
200
300
400
500
600
0
200
400
600
800
1000
1200
Picea abies (sample 14)
δ13C
(‰)
-23.5
-23.2
-22.9
-22.6
-22.3
Rad
ialc
ells
ize
(mm
)
0
10
20
30
40
50
Cel
lwal
l thi
ckne
ss(µ
m)
0
1
2
3
4
5
6
7
Date
Cel
lwal
lare
a(µ
m2 )
Lum
enar
ea(µ
m2 )
Woo
dde
nsity
(mg
cm-3)
Radial cell size
Cell wall thickness
Lumen area
Wood density
Cell wall area
May Jun Jul Aug Sep2001
Fig. 2 Example of anatomical,
density and isotopic
measurements in a spruce tree
ring (Renon, Italy; Picea abies,
tree number 14): radial cell size,
cell-wall thickness, lumen area,
cell-wall area, wood density.
The laser shots for isotope
determination can be seen as
vertical bars
734 Oecologia (2009) 161:729–745
123
Table 3 Statistical parameters of tree-ring width and maximum density chronologies for conifers from the studied sites
Species Parameter Site Number
of samples
Standard deviation of
relative TRW and MaxD
Inter-serial correlation
(average ± SD)
Common variance
(first Eigen-vector)
First-order
autocorrelation
Picea abies TRW REN 13 0.10 0.32 ± 0.20 32.3 0.75
TRW HAIa 14 0.20 0.68 ± 0.21 61.2 0.54
TRW FLAa 13 0.19 0.71 ± 0.46 61.1 0.76
MaxD REN 8 0.06 0.60 ± 0.06 54.2 -0.16
MaxD HAI 14 0.07 0.74 ± 0.08 69.7 0.03
MaxD FlA 13 0.07 0.64 ± 0.09 – –
Pinus cembra TRW REN 8 0.16 0.28 ± 0.18 35.2 0.72
Pinus sylvestris TRW FLA 18 0.17 0.58 ± 0.23 55.3 0.58
Pinus cembra MaxD REN 3b 0.06 0.30 ± 0.06 37.7 0.19
Pinus sylvestris MaxD FLA 18 0.07 0.67 ± 0.09 55.2 0.21
TRW, Tree-ring width; MaxD, maximum diameter; SD, standard deviation; REN, Renon (Italy), an alpine forest in North Italy; HAI, Hainich, a
lowland forest in Germany; FLA, a boreal forest in Northern Sweden (Flakaliden, FLA)a Data from ITRDB (FH Schweingruber)b Number of samples is not enough for significance
Maximum wood density
1900 1920 1940 1960 1980 20000.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
1750 1800 1850 1900 1950
Sta
ndar
dize
d m
axim
um d
ensi
ty
Years1900 1920 1940 1960 1980
Tree ring width
1900 1920 1940 1960 1980 2000
Renon (Italy)
1750 1800 1850 1900 1950
Sta
ndar
dize
d tr
ee r
ing
wid
th
Hainich (Germany)
Years1900 1920 1940 1960
Flakaliden (Sweden)
Pinus cembra
Picea abiesPinus cembra
Picea abies
Picea abies Picea abies
Picea abies
Picea abies
Pinus sylvestris
Pinus sylvestris
Fig. 3 Standardized site tree-
ring width and maximum
density chronologies for Piceaabies and Pinus cembra. The
original data were standardized
by correcting for an age-
dependent trend
Oecologia (2009) 161:729–745 735
123
inter-serial correlation was higher for tree-ring width than
for maximum wood density, and it was highest at FLA and
lowest at REN. Thus, there was relatively little variation
between trees at FLA and HAI, but a large variation
between trees at REN. Also, the common variance of the
maximum density and tree-ring width was higher at FLA
and HAI than at REN, indicating a weaker climatic signal
at REN.
The first-order autocorrelation in tree-ring width, which
is an indicator of the carry-over effect from one year to the
next, is very high at all study sites. In fact, 29–58% of tree
ring-width in a current year is explained by the growth rate
in the previous year (Table 3). The autocorrelation in the
tree rings is much larger than the autocorrelation of cli-
matic parameters, which was 0.06, 0.08 and 0.16 for tem-
perature, and 0.10, -0.12 and -0.06 for precipitation at the
REN, HAI and FLA site, respectively (data not shown).
This result indicates that only 0.3–2% of the climatic
conditions in one year are explained by the climatic con-
ditions in the previous year.
The correlation of long-term chronologies of tree growth
and wood density with temperature and precipitation (time
series 70–100 years) shows no general pattern between tree-
ring width or maximum wood density and monthly mean
values for precipitation and temperature (Fig. 4). Site-
specific conditions are important. For example, late season
temperature (August, September, October) correlated with
tree-ring width at the alpine REN and boreal FLA sites, but
the early season temperature (May, June, July) was signifi-
cant at the lowland HAI site. Precipitation had no significant
effect on tree-ring width at the alpine site, but late season
precipitation correlated with tree-ring width at HAI, and
early season precipitation was significant at FLA despite
snow melt. In general, there is a stronger climatic effect at
higher latitudes (Shiyatov 1986; Vaganov et al. 1996; Briffa
et al. 1998) than at the alpine site. The climate signal was
not as strongly expressed as that observed for maritime
Mediterranean sites (Gonzales and Eckstein 2003).
In summary, we found that the long-term correlations
between climate and tree-ring width and wood density are
not very strong and that they are both site- and species-
specific. For tree-ring width, late season temperatures (but
not precipitation) are significant at REN, and spring tem-
peratures and precipitations are significant at HAI and FLA.
Extreme events of temperature and precipitation, such as in
the dry year of 2003 (see Fig. 7), seem to be important.
Relations between isotopic composition and cell lumen
and tree-ring width
All sites show a positive correlation between the average
value of d13C within a tree ring and the width of tree rings
(Fig. 5). This was unexpected because a more positive
d13C value would indicate an increasing stomatal limitation
of photosynthesis with increasing ring width. The main
differences between sites are the intercept values and a
lower slope at the REN site. In narrow rings, the temperate
and the boreal site are more similar than the alpine and the
boreal site. The high d13C values at REN could indicate
water stress at the alpine site. This was not apparent in the
correlations with long-term chronologies (Fig. 4).
The variations in d13C and wood anatomy within a tree
ring were investigated for the REN site (Fig. 6) by calcu-
lating the deviation of d13C from the average value for a
given tree ring. Cell lumen area was used as a parameter
because it is positively correlated with tracheid diameter
(overall correlation 0.86, P \ 0.0001), and it is negatively
correlated with CWT (overall correlation -0.48,
P \ 0.01), cell-wall area and density (Lambers et al. 1998;
Hubbard et al. 2001).
Cell-lumen area shows the highest values at the begin-
ning of the growing season and continually decreases later
in the season. The decrease is steeper in narrow than in
wide rings. The maximum values of cell lumen area are
slightly higher for spruce (1063–1225 lm2) than for pine
(956–1058 lm2). There is no obvious change at the border
of early and late wood.
The seasonal changes of d13C with tree-ring width can
be seen to differ from the seasonal changes in cell-lumen
area (Fig. 6). At the beginning of the season, average d13C
was 0.4–0.8% lower than the tree-ring average; it then
increased with progressing tree-ring formation (see Fig. 5),
but decreased again past a maximum. This decrease may
only be small in narrow rings, but it is substantial in wide
tree rings. Thus, in wide tree rings, the late wood d13C may
be lower than in the early wood, while in narrow rings d13C
in late wood is higher than in early wood of the same ring.
The average amplitude of the intra-annual variation of d13C
is about 0.8–1.6% depending on years of growth.
The changes in d13C within the tree rings (see Fig. 6) do
not correlate with the observed long-term effect of climate
on tree-ring width (Fig. 4) and may reflect seasonal chan-
ges in the contribution of storage products for tree-ring
growth.
The seasonal and inter-annual variation
of d13C in spruce
In this section, we depict examples of the seasonal changes
in d13C in fast-, average- and slow-growing trees of Picea
abies with the aim of demonstrating the tree-by-tree vari-
ability as well as inter-annual variability for the three sites
(Fig. 7). The different numbers of years in Fig. 7 are due to
the sampling procedure, where cores of different lengths
were taken from fast- and slow-growing trees. Thus, not all
cores reach the same depth from the cambium.
736 Oecologia (2009) 161:729–745
123
Figure 7 shows that the variation in d13C within years
(intra-annual variation) is highly variable between fast- and
slow-growing trees. d13C was either constant, increased or
decreased over time, with no consistent relations.
In fast-growing trees, the d13C of early wood was on
average the same for all three sites (24.9 ± 0.6%). The
seasonal course of d13C in fast-growing trees at REN
showed an autumn decrease, even in the dry years of 2001
and 2003. At HAI, dry years (2001 and 2003) differ from
wet years in that there was an increase in autumn d13C in
the dry years. At FLA also, there may be an increase in
d13C in the autumn of wet years, and there was no d13C
signal left in fast-growing trees of FLA during the very dry
year of 2002.
Slow-growing trees d13C showed a more consistent
pattern than fast-growing trees. At REN, d13C generally
increased as the season progressed, which was opposite to
the seasonal pattern observed in fast-growing trees.
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Cor
rela
tion
coef
ficie
nts
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Tree ring width
Cor
rela
tion
coef
ficie
nts
Maximum wood density
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Hainich (Germany)
Flakaliden (Sweden)
Picea abies Picea abies
Picea abies Picea abies
Months
Renon (Italy)
MonthsOct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Pinus cembra Pinus cembra
* *
Picea abies Picea abies
* *
1950-1998
**
*
**
**
*
*
**
**
*
*
* * * * *
1957-2003
1901-1998
T T
T
T
T
T
T
TP
P
P
P
PP
P
P
Fig. 4 Correlation coefficients
of annual tree-ring and
maximum density chronologies
with climatic data: Linesindicate correlations with
monthly mean temperature,
columns indicate correlations
with monthly precipitation.
Asterisks indicate a significant
correlation (P \ 0.01)
Oecologia (2009) 161:729–745 737
123
At HAI, the effect of a dry year was not apparent in slow-
growing trees, while at FLA, the very dry year of 2002 did
not leave a signal in slow-growing trees, but the lesser dry
year of 1999 and 1997 resulted in a increase in d13C.
The highest values of d13C were found at REN for a tree
of average growth rate (-22.8%) with very little inter-
annual and seasonal variation. The d13C values were gen-
erally lower in slow-growing trees than in fast-growing
ones. The lowest values were observed in slow-growing
trees of HAI and FLA (-27%). It should be pointed out
that all trees were dominant in height and that none of the
trees was a suppressed tree of the lower canopy.
There was a remarkable stepwise change in some trees
in terms of d13C between late wood and early wood of the
next year. A decrease in d13C across the tree ring boarder
was observed at REN, where the lowest d13C values were
observed in autumn. This result contrasts with those
obtained at the HAI and FLA sites, where the d13C values
of the early wood were generally lower than those of the
late wood in autumn.
In order to obtain insight into the processes which link
the late wood of one year with the early wood of the next
year, we assessed whether the d13C in ray cells differed
from the d13C in xylem cells (Table 4). The expectation
was that ray cells, which contain the stored carbohydrates
in conifers, would contain an additional isotope signal; i.e.
the hypothesis was that ray cells would be heavier that the
associated tracheids. However, at all sites, the d13C in 1- to
3-year-old ray cells was 0.1–0.2% lower than that of the
associated xylem. This difference disappeared after
4 years, i.e. ray cells were the same as xylem cells. A lower
d13C value could be the result of a continuation of lignin
incorporation into ray cells (FH Schweingruber, personal
communication), which would lower the d13C of ray cells.
Obviously, the bulk measurement of the whole cell is not
fully conclusive and does not show the effects of carbo-
hydrate turnover.
Independent of sites, yearly climate and growth rates
(Fig. 8), we found a close correlation between the yearly
average d13C of a tree ring in a given year (t) and the
Tree ring width (mm)0.0 0.5 1.0 1.5 2.0 2.5 3.0
δ13C
(‰
)
-28
-27
-26
-25
-24
-23
-22
Flakaliden
Hainich
Renon
Renon: y=-24.33+0.39x, r2=0.053, p=0.165
Hainich: y=-26.16+0.73x, r2=0.221, p=0.024
Flakaliden: y=-26.97+0.81x, r2=0.375, p=0.003
Fig. 5 Correlation between tree-ring width and average value of d13C
within tree rings for small (\3 mm) tree rings. The average d13C
value represents the numerical average of the individual measure-
ments without taking changes in density into account. The higher
density of late wood affects only 10–20% of the d13C data
Picea abies
0 1 2 3 4 0 1 2 3 4
0 1 2 3 4 0 1 2 3 4
0
200
400
600
800
1000
1200
1400narrow; y=916.90-0.128x-0.0011x2, r2=0.785
middle; y=1027.90-0.304x-0.0002x2, r2=0.745
wide; y=943.14-0.032x-0.0002x2, r2=0.788
very wide; y=890.07-0.043x-3.99e-5x2, r2=0.890
Pinus cembra
Cel
l lum
en a
rea
(µm
2 )
0
200
400
600
800
1000
1200
1400narrow; y=872.89-0.705x-7.59e-5x2, r2=0.620
middle; y=893.17-0.176x-0.0003x2, r2=0.777
wide; y=947.58-0.164x-9.58e-5x2, r2=0.865
Distance from tree ring border (mm)
narrow; y=-0.456+0.0015x-8.78e-7x2, r2=0.327
middle; y=-0.461+0.0013x-7.04e-7x2, r2=0.374
wide; y=-0.207+0.0008x-4.04e-7x2, r2=0.374
very wide; y=-0.125+0.0006x-2.05e-7x2, r2=0.654
δ13C
-δ13
Cav
erag
e (‰
)
Cel
l lum
en a
rea
(µm
2 )δ13
C-δ
13C
aver
age
(‰)
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
narrow; y=-0.712+0.0032x-2.50e-6x2, r2=0.596
middle; y=-0.680+0.002x-1.08e-6x2, r2=0.484
wide; y=-0.849+0.0019x-7.62e-7x2, r2=0.533
Renon RenonFig. 6 Relations between cell
lumen area and d13C with tree-
ring width as expressed by
distance from the tree ring
boundary. The Renon site was
taken as an example
738 Oecologia (2009) 161:729–745
123
average d13C in the immediately preceding year (t-1). At
low d13C values, there was a tendency for the yearly
average d13C of early wood to be higher than that of late
wood; in contrast, at high d13C values, the yearly average
d13C in early wood would tend to be lower than that in late
wood.
It was not possible to compare d13C in wood with eddy
flux data at all sites because the ecosystem flux at the
alpine site REN is also determined by the activity of the
grass cover on the forest floor, while the stand was too
small at the lowland HAI site for eddy flux measurements.
At the boreal site FLA, it was only possible to make a
comparison for 2001 and 2002 before the eddy flux mea-
surements terminated. The synchronization with growth
was made on the basis of independent dendrometer mea-
surements (Sune Linder, personal communication). We
used gross primary productivity and canopy conductance to
derive a canopy-scale ratio of the intercellular to ambient
CO2 concentration (ci/ca), which is not equal to leaf–level
ci/ca, but rather a integrative estimate of ci/ca across the
entire canopy confounded by other effects, such as surface
evaporation and mixing sun and shade leaves. Figure 9
shows an increase in d13C early in the season in both years,
when canopy-scale ci/ca was increasing. Thus, the initial
increase in d13C is in contrast to the expectation that gas
exchange parameters determine the isotope composition in
wood. Also, the d13C in the following year was lower or
similar to that in the past autumn, which was contrasted by
much lower ci/ca values. Thus, gross primary productivity,
canopy conductance and canopy-scale ci/ca do not explain
the observed variation in d13C in wood.
The information obtained during our study, which is
based on a total of three sites, 22 years of wood analysis
and 760 individual observations, was analyzed with linear
mixed-effects models to explore relationships between
d13C, wood anatomy, yearly climate and their dependency
on different study sites (Table 5). Cell-wall area was
related to cell lumen in early and late wood, and d13C
correlated with cell-wall area and lumen in early and late
wood even if the significant interaction terms with site
identity indicated differences in these relationships among
sites. The relationships to climate parameters are compli-
cated and mostly site-specific. Tree-ring width correlated
with early season precipitation and late season temperature
(mainly an effect of HAI). Vapor pressure deficit (VPD)
and the total incoming radiation (R) were highly significant
variables to explain variation in tree-ring width although
their effects varied to some degree among sites as well.
d13C varied among sites, but it generally did not correlate
with precipitation or temperature. However, despite site-
specific differences, d13C was highly significantly related
to total radiation, VPD and the length of the growing
HAI-Sp3015
0
HAI-Sp3014
REN-Sp8
024681012 24681012 24681012
2000 2001 2002 2003
REN-Sp4
Distance from cambium (mm)
2002
2003
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
1998 1999 2001 2002 20032000REN-Sp14 FLA-Sp9
1999 2000 2001 2002 2003
FLA-Sp3
δ13C
(‰
)
1998
1999
2000
1997
2001
2002
FLA-Sp1
0
0024681012 24681012 24681012 0
0024681012 24681012 24681012 0
-28
-27
-26
-25
-24
-23
-22
-28
-27
-26
-25
-24
-23
-22
-28
-27
-26
-25
-24
-23
-22
20032002200120001999
2000 2001 2002
2003
2004
HAI-Sp30162000 2001 2002 2003 2004
2004200320022001
earlywoodlatewood
tree
with
rap
id g
row
thav
erag
e gr
owth
slow
gro
wth
Fig. 7 Seasonal courses of d13C during consecutive years, taking a
fast-growing tree, a tree with average growth rate and a slow-growing
tree as an example. Closed symbols indicate early wood, open
symbols indicate late wood. Columns: left Renon (REN), middleHainich (HAI), right Flakaliden (FLA)
Oecologia (2009) 161:729–745 739
123
season. The correlation between early wood of one year
versus late wood of the immediately preceding year was
highly significant.
Figure 10 summarizes the yearly averages of all data
points in terms of yearly d13C and tree-ring width. There is
a strong latitudinal effect of d13C, which was highest at
REN and lowest at FLA. Although REN receives 20–40%
more solar radiation than HAI and FLA, respectively, tree-
ring width at REN varies mostly between that at HAI and
FLA. The HAI site receives only 30% more solar radiation
than FLA, but the tree-ring width was found to be twofold
larger. Despite high rainfall, REN appears to be a water-
stressed site due to the special soil conditions, but water
stress alone cannot explain the differences. Tree-ring width
is lowest in Flakaliden, despite low d13C. There are obvi-
ously additional factors which determine tree-ring width
and d13C, one of which is plant nutrition. FLA is nitrogen
deficient (Table 1), while the nitrogen concentrations of
needles were the highest at REN. High nitrogen availability
is expected to increase carboxylation capacity and to
amplify water stress, leading to an increase of d13C in leaf
and wood (Hogberg et al. 1993; Livingston et al. 1999) and
thus explain part of our observation.
Discussion
This study was undertaken to investigate correlations
between climate and tree-ring width and anatomy at long
and short time scales and to explore whether d13C could be
used as an additional parameter to interpret tree-ring chro-
nologies. We also wanted to know if there are carry-over
effects from one year to the next that affect d13C and
growth. We investigated these questions by (1) studying
climate correlations of long-term tree ring chronologies, (2)
investigating in detail short-term seasonal observations
using wood samples which were taken from forest sites
ranging between boreal Scandinavia and alpine North Italy
where trees grew in competitive interaction as forest stands.
In the long time series, temperature and precipitation
correlated with tree-ring width and wood density, but these
correlations were highly site and species specific, with
spring-time temperatures and precipitations seeming to be
most important. In contrast to the long-term correlations, at
short time scales, we found only poor correlations between
tree-ring width and temperature or precipitation. However,
there was a highly significant correlation between tree-ring
width, d13C and lumen area and the length of the growing
season, total net radiation and VPD. Our investigation also
shows that there were large differences between fast- and
slow-growing trees even though all trees were dominant
canopy trees in which the d13C signal was not affected by
soil respiration. It is quite likely that the main driver for
d13C could be VPD. We were unable to test for effects of
soil moisture because these data were not available as time
series for all sites.
Our initial hypothesis was that the seasonal course of
d13C would allow an identification of dry years (i.e. years
Table 4 Difference between d13C of xylem and ray cells
Year Early (EW)-Late
wood (LW)
Hainich
d13C-xylem
Hainich
d13C-ray
Hainich d13C
xylem-ray
Renon
d13Cxylem
Renon
d13C-ray
Renon d13C
xylem-ray
Flakaliden d13C
xylem-ray
2001 EW -25.51 ? 0.42 -25.98 ? 0.42 -0.01 ? 0.02
LW -25.44 ? 0.79 -25.54 ? 0.80 0.11 ? 0.14
2002 EW -25.62 ? 0.56 -25.69 ? 0.58 0.07 ? 0.17 0.16
LW -25.84 ? 0.37 -25.95 ? 0.31 0.11 ? 0.12 -24.66 ? 1.20 -24.68 ? 1.20 0.01 ± 0.05 0.21
2003 EW -25.37 ? 0.62 -25.60 ? 0.68 0.23 ? 0.13 -24.36 ? 0.85 -24.48 ? 0.83 0.12 ± 0.13
LW -23.91 ? 1.39 -24.10 ? 1.34 0.18 ? 0.12 -24.31 ? 1.35 -24.40 ? 1.35 0.08 ? 0.15
2004 EW -26.11 ? 0.61 -26.26 ? 0.69 0.15 ? 0.12
LW -25.44 ? 0.79 -25.54 ? 0.80 0.09 ? 0.07
δ13C
ew (
curr
ent y
ear) (
‰)
δ13Clw (previous year)
(‰)
Picea abies
-28 -27 -26 -25 -24 -23 -22-28
-27
-26
-25
-24
-23
-22
RenonHainichFlakaliden
y=-5.83+0.78x, r2=0.626, p=0.0000
1:1
Fig. 8 Relation between d13C in late wood of previous tree ring and
in early wood of current tree ring for spruce. The thin line shows the
1:1 relation. The regression lines are calculated by reduced axis
regression
740 Oecologia (2009) 161:729–745
123
with below-average rainfall during the growing season) as
had been described by Ferrio et al. (2003) in Pinus
halepensis, Kagawa et al. (2003) in Larix gmelinii or
Adams and Kolb (2004) in Pinus ponderosa. We observed
a d13C response in a fast-growing tree of HAI during the
dry year of 2003; however, the effect disappeared in slow-
growing dominant trees, and the d13C increase did not
increase at the other sites during years with a below-
average rainfall. To the contrary, d13C increased even in
wet years at FLA. d13C did not respond at REN. There is
clearly an absence of a general and regular response of
tree-ring width and d13C to precipitation. The site effects
and the effects of slow- versus fast-growing trees override
the precipitation effect.
The seasonal increase in d13C at FLA was not caused by
drought. Rainfall at FLA is generally lower in August, but
this did not affect canopy-scale ci/ca, and the d13C of wood
changed independently of canopy-scale ci/ca. We hypoth-
esize that the increase in d13C in autumn is caused by the
reserve formation of products which are 13C-enriched and
which then result in heavy early season wood.
In contrast, at REN, d13C decreased in fast-growing
trees but was constant or increased in slow-growing trees.
Thus, the seasonal pattern is more complex at REN than at
the FLA site and is not simply related to one climatic
factor. However, slow- and fast-growing trees could
represent different rooting depths at this alpine site, which
consists of very stony soils. We cannot rule out that slow-
rowing trees at the REN site experience water stress, but
this was not apparent in the seasonal pattern of fast-
growing trees.
At HAI, d13C generally increased during the year, and
this increase occurred concurrently with a 20% decrease in
precipitation and a 40% decrease in solar radiation
(Table 2). At this site, the effect was most pronounced in
fast-growing trees and was not observable in slow-growing
trees, which is opposite to the observation at REN.
In summary, TRW is correlated with early and late
season precipitation in both the long-term and short-term
time series. However, there was no simple relationship
between d13C and precipitation. There was a significant
correlation with VPD, which is expected to close stomata,
and with radiation, which is expected to open stomata. At
FLA, the ci/ca remained constant during the autumn with
low rainfall but relatively high radiation. A comparison of
the three sites indicates that an interaction with nutrition is
likely. The site with the lowest nitrogen concentrations in
the needles had the lowest d13C values. This result corre-
sponds well with findings from nitrogen fertilization
experiments in which low nitrogen availability resulted in
low d13C in the needles of Pinus sylvestris L. (Hogberg
et al. 1993; Betson et al. 2007), Picea abies Karst.
gc (
mm
olm
-2 s
-1)
0
200
400
600
800
ci/c
a
0.4
0.6
0.8
1.0
δ13C
(‰)
-27.0
-26.0
-25.0
-24.0
-23.0
-22.0
ci/ca
earlywood δ13C
latewood δ13C
Flakaliden
GP
P(µ
mol
m-2 s
-1)
-5
0
5
10
15
20
SP1
SP9
SP3
SP1
SP9
SP3
J F M A M J J A S O N D J F M A M J J A S O N D
2001 2002Date
Fig. 9 Seasonal course of gross
primary productivity (GPP) as
measure of photosynthesis,
canopy conductance (gc), the
ratio of canopy-scale mesophyll
internal and atmospheric CO2
concentration (ci/ca) estimated
based on eddy covariance flux
measurements, and d13C in
wood for the boreal site at
Flakaliden in 2001 and 2002.
Closed symbols of d13C indicate
early wood, open symbolsindicate late wood
Oecologia (2009) 161:729–745 741
123
(Hogberg et al. 1993) and Picea glauca (Moench) Voss
(Livingston et al. 1999). Possible mechanisms include
changes in carboxylation capacity (Livingston et al. 1999)
and an increase in susceptibility to climate-induced water
stress with increasing nitrogen availability (Hogberg et al.
1993; Betson et al. 2007).
The most striking observation is the close correlation
between the early wood of a given year and the late wood
of immediately preceding year, which indicates a carry-
over effect of stored products (Helle and Schleser 2004;
Skomarkova et al. 2006). Thus, our hypothesis was that the
d13C in one year is related to the d13C in the previous year
Table 5 Summary of linear mixed model analyses of the relationships between wood anatomy and carbon isotope ratios and their dependency
on climatic variables
Relationship Tested variablesa Site Variable Interaction site
9 variable
All data d13C vs. anatomy CWA vs. LUM 4.93 130.29*** 21.35***
d13C vs. CWA 6.00* 17.45*** 14.46***
d13C vs. LUM 6.00* 4.62* 64.15***
Early wood CWA vs. LUM 4.03 265.03*** 40.46***
d13C vs. CWA 6.94* 16.19*** 16.25***
d13C vs. LUM 6.94* 29.97*** 24.63***
Late wood CWA vs. LUM 15.60*** 60.51*** 14.86***
d13C vs. CWA 2.95 15.24*** 4.14
d13C vs. LUM 2.95 6.23* 8.27*
TRW vs. climate TRW vs. P-MJJ 3.23 7.38** 16.86***
TRW vs. P-ASO 3.23 0.30 32.60***
TRW vs. T-MJJ 3.23 1.07 1.64
TRW vs. T-ASO 3.23 9.40** 2.85
TRW vs. R-MJJ 3.23 21.67*** 6.04*
TRW vs. R-ASO 3.23 16.51*** 9.61**
TRW vs. VPD-MJJ 3.23 15.01*** 11.09**
TRW vs. VPD-ASO 3.23 18.09*** 5.81
d13C vs. TRW d13C vs.TRW 6.00* 3.19 1.83
Early wood d13C vs. TRW 6.94* 2.75 1.02
Late wood d13C vs. TRW 2.95 4.67* 4.31
TRW vs. growing season TRW vs. length 3.23 11.09*** 9.23**
LUM vs. growing season LUM vs. length 9.43** 594.07*** 0.45
LUM vs. Radiation LUM vs. R-MJJ 9.43** 1089.27*** 1.71
LUM vs. R-ASO 9.43** 1089.11*** 1.12
CWA vs. growing season CWA vs. length 4.93 18.92*** 5.93
d13C vs. growing season d13C vs. length 6.00* 18.92*** 2.14
d13C vs. climate d13C vs. P-MJJ 6.00* \0.01 14.23***
d13C vs. P-ASO 6.00* 2.43 8.55*
d13C vs. T-MJJ 6.00* 0.44 0.31
d13C vs. T-ASO 6.00* 1.08 9.76**
d13C vs. R-MJJ 6.00* 45.01*** 14.58***
d13C vs. R-ASO 6.00* 40.08*** 1.95
d13C vs. VPD-MJJ 6.00* 47.52*** 7.02*
d13C vs. VPD-ASO 6.00* 43.60*** 9.61**
Early vs. late wood preceding year * *** NS
*, **, *** Significant at the 5, 1 and 0.1% level, respectively
Models were fitted by the stepwise inclusion of variables. Listed are the values of likelihood ratio statistics that were applied to assess model
improvement and the statistical significance of the variablesa Variables: TRW, tree-ring width; CWA, cell-wall area; LUM, cell lumen; P, precipitation; T, temperature; R, incoming radiation; VPD, vapor
pressure deficit; length, length of the growing season (days with average temperature [5�C); MJJ, early season (May, June, July); ASO, late
season (August, Spetember, October)
742 Oecologia (2009) 161:729–745
123
through storage products. The use of stored carbohydrates
and lipids for growth in consecutive years has been con-
firmed by Keel et al. (2007) in a permanent 13C-fumigation
experiment of deciduous trees. The accumulation of
reserves involves an accumulation of carbohydrates in the
autumn and a depletion of carbohydrates in the spring. It
also involves an accumulation of lipids during the summer,
which are metabolized in spruce in the autumn (Hoell
1985). However, other tree species, such as Larix, may
accumulate lipids during the winter (Sudachkova et al.
2004). The effect of storage has been demonstrated in
deciduous trees (Helle and Schleser 2004) which have a
larger number of wood parenchyma cells than conifers
(Schweingruber et al. 2007). One would not expect major
storage pools in conifers, which have only a few ray cells.
However, Hoell (1985) did demonstrate the turnover of
soluble carbohydrates in conifers, and Von Felten et al.
(2007) estimated that 42% of new wood in Larix originates
from storage products.
Direct proof that stored carbohydrates of one year con-
tribute to the growth of the next year was presented for
conifers by Hansen and Beck (1994), who pulse-labeled
young Pinus sylvestris trees with 14C in the field and
demonstrated the incorporation of 14C into wood of the
following year. The process was biochemically further
complicated by the fact that starch was transformed into
various soluble carbohydrates, such as maltotriose, gal-
actose, arabinose and rhamnose in the winter as a mecha-
nism of cold hardiness (Hansen and Beck 1994). Thus, the
breakdown of starch in wood results in the production of
triose-phosphates, which results again in a 13C depletion
step (Gleixner et al. 1998). Obviously, it is also not pos-
sible to demonstrate fluxes of 13C in retrospect using tree
rings and ray cells.
Gessler et al. (2008) measured d13C real time in the
phloem sap of Rhizinus and found an increase as well as a
decrease in d13C in the phloem sap compared to the
immediate photosynthetic product, depending on whether
the phloem sugars were derived from sucrose or from
starch or from sucrose passing through the cell metabolic
pathways, including the formation of triose-phosphates. No
fractionation was found during phloem transport. However,
this experiment did not include the formation of wood in a
seasonal climate, which may be supported by carbohy-
drates from ray cells in the wood and not involve phloem
transport. This situation is different from that of continually
fast-growing trees (Pinus radiata: Walcroft et al. 1997;
Eucalyptus globulus: Pate and Arthur 1998) where a cor-
relation between the phloem sap isotope signal and d13C in
wood has been demonstrated. During the loading of storage
cells, carbohydrates move from the phloem via the cam-
bium and the xylem initials into the ray cells of existing
wood. If wood formation has priority over storage, the
storage products would be expected to become relatively
enriched in 13C (increasing d13C). However, if storage and
cold hardiness has priority, then d13C could also be
depleted (decreasing d13C). In contrast, during carbohy-
drate mobilization in the spring, carbohydrates move from
ray cells into the cambium without entering the phloem. It
would be the mixture of the enriched mobilized carbohy-
drates and the phloem-derived carbohydrates from spring
photosynthesis which determines the d13C of wood. An
additional important step of discrimination takes place
during wood respiration due to the breakdown of carbo-
hydrates into triose-phosphates. However, respiratory CO2
is 13C enriched (Brandes et al. 2006), which does not
explain the enrichment of d13C in stems (Damesin and
Lelarge 2003).
In our study, a stepwise change between late and early
wood of consecutive years was observed, but this stepwise
change was not always in the same direction, as was pre-
dicted by Helle and Schleser (2004). We observed all
possible permutations of alternative responses depending
on the site, species and climate. Thus, the linkage between
the d13C of wood and climate is not constant. There are
many ways which link d13C in wood, storage and climate
and which govern the relation between tree ring width and
d13C. Thus, based on our results, we conclude that tree-ring
width is not consistently related to d13C in seasonal
climates, and both are only indirectly linked to climate.
Tre
e rin
g w
idth
(m
m)
0.5
1.0
1.5
2.0
2.5
3.0
3.5
FLA
HAI
REN
Picea abies
Year1999 2000 2001 2002 20031999 2000 2001 2002 2003
δ13C
(‰)
-26.5
-26.0
-25.5
-25.0
-24.5
-24.0
-23.5
FLA
HAI
REN
Fig. 10 Inter-annual variations
in absolute values of d13C in
tree rings of spruce (all three
studied sites) during the last 5
years (1999–2003). Verticallines show the standard
deviations
Oecologia (2009) 161:729–745 743
123
Conclusion
The carry-over effect, based on correlations between the
d13C of early wood and the d13C of late wood of the
immediately preceding year is significant in conifers.
However, depending on sites and specific meteorological
conditions, other effects, such as whole-wood composition,
the isotope signal of storage products or a dry autumn
leading to dry spring result in additional variation.
The use of a 5�C average temperature as a proxy for the
start of the growing season appears to be robust for boreal,
temperate and Mediterranean sites. The correlations in this
study are insensitive to this proxy, but more elaborate
phasing studies are needed in the future.
Acknowledgments This work was supported by Alexander von
Humboldt (Research Award 2003 for E. Vaganov) and the Russian
Foundation of Basic Research (RFBR-05-04-48069). We thank
Alessandro Cescatti, Leonardo Montagnani, Stefano Minerbi and
Claudio Mutinelli for providing the climate and nitrogen data for
Renon, Sune Linder for dendrometer data, and Anders Lindroth for
eddy flux data of the Flakaliden site. We thank Gerd Gleixner
for discussion of this manuscript. We also like to thank Annett
Boerner for the artwork and Jens Schumacher for advice on statistical
analyses.
Open Access This article is distributed under the terms of the
Creative Commons Attribution Noncommercial License which per-
mits any noncommercial use, distribution, and reproduction in any
medium, provided the original author(s) and source are credited.
References
Adams HD, Kolb TE (2004) Drought responses of conifers in ecotone
forests of northern Arizona: tree ring growth and leaf d13C.
Oecologia 140:217–225
Bates D, Sakar D (2006) LME4: linear mixed-effects models using S4
classes. R package version 0.995-2. Available at:
http://www.r-project.org
Betson NR, Johannisson C, Lofvenius MO, Grip H, Granstrom A,
Hogberg P (2007) Variation in the delta C-13 of foliage of Pinussylvestris L. in relation to climate and additions of nitrogen:
analysis of a 32-year chronology. Glob Chang Biol 13:2317–2328
Brandes E, Kodama N, Whittaker K, Weston C, Rennenberg H, Keitel
C, Adams MA, Gessler A (2006) Short-term variation in the
isotopic composition of organic matter allocation from the leaves
to the stem of Pinus sylvestris: effects of photosynthetic and
postphotosynthetic carbon isotope fractionation. Glob Chang
Biol 12:1922–1939
Briffa KR, Schweingruber FH, Jones PD, Osborn TJ, Shiyatov SG,
Vaganov EA (1998) Reduced sensitivity of recent tree growth to
temperature at high northern latitudes. Nature 391:678–682
Brugnoli E, Farquhar GD (2000) Photosynthetic fractionation of
carbon isotopes. In: Leegood RC, Sharkey TD, von Caemmerer
T (eds) Advances in photosynthesis: physiology and metabolism,
vol 9. Springer, Berlin Heidelberg New York, pp 399–434
Cescatti A, Marcolla B (2004) Drag coefficient and turbulence
intensity in conifer canopies. Agric For Meteorol 121:197–206
Cook ER, Kairiuktis LA (eds) (1990) Methods of dendrochronology.
Application in the environmental sciences. Kluwer, Dordrecht
Cook ER, Peters K (1981) The smoothing spline: a new approach to
standardizing forest interior tree-ring width series for dendrocli-
matic studies. Tree Ring Bull 41:45–53
Cook ER, Briffa KR, Shiyatov SG, Mazepa VS (1990) Tree-ring
standardization and growth-trend estimation. In: Cook ER,
Kairiukstis LA (eds) Methods of dendrochronology. Application
in the environmental sciences. Kluwer, Dordrecht, pp 104–123
Damesin C, Lelarge C (2003) Carbon isotope composition of current-
year shoots from Fagus sylvatica in relation to growth,
respiration and use of reserves. Plant Cell Environ 26:207–219
Duquesnay A, Breda N, Stievenard M, Duponuey JL (1998) Changes
of tree ring delta C-13 and water-use efficiency of beech (Fagussylvatica L.) in north-eastern France during the past century.
Plant Cell Environ 21:565–572
Eglin T, Maunoury-Danger F, Fresnau C, Lelarge C, Pollet B,
Lapierre C, Francois C, Damesin C (2008) Biochemical
composition is not the main factor influencing variability in
carbon isotope composition of tree rings. Tree Physiol 28:1619–
1629
Farquhar GD, Ehleringer JR, Hubik KJ (1989) Carbon isotope
discrimination and photosynthesis. Annu Rev Plant Physiol Plant
Mol Biol 40:503–537
Ferrio JP, Florit A, Vega A, Serrano L, Voltas J (2003) Delta c-13 and
tree-ring width reflect drought responses in Quercus ilex and
Pinus halepensis. Oecologia 137:512–518
Francey RJ, Farquhar GD (1982) An explanation of C-13/C-12
variations in tree rings. Nature 297:28–31
Fritts HC (1966) Growth-rings of trees: their correlation with climate.
Science 154:873–879
Fritts HC (1976) Tree-rings and climate. Academic Press, London
Furst GG (1979) Methods of anatomical and histochemical research
of plant tissue (in Russian). Nauka, Moscow
Gessler A, Tcherkez G, Peuke AD, Ghashghaie J, Farquhar GD
(2008) Experimental evidence for diel variations of the carbon
isotope composition in leaf, stem and ploem sap organic matter
in Ricinus communis. Plant Cell Environ 31:941–953
Gleixner G, Scrimgeour C, Schmidt HL, Viola R (1998) Stable
isotope distribution in major metabolites of source and sink
organs of Solanum tuberosum L.: a powerful tool in the study of
metabolic partitioning in intact plants. Planta 207:241–245
Gonzales IG, Eckstein D (2003) Climate signal of earlywood vessels
of oak in a maritime climate. Tree Physiol 23:497–504
Hansen J, Beck E (1994) Seasonal changes in the utilization and
turnover of assimilation products in 8-year-old Scots pine (Pinussylvestris L.) trees. Trees 8:172–182
Harlow BA, Marshall JD, Robinson AP (2006) A multi-species
comparison of d13C from whole wood, extractive-free wood and
holocellulose. Tree Physiol 26:767–774
Helle G, Schleser GH (2004) Beyond CO2-fixation by Rubisco—an
interpretation of 13C/12C variations in tree rings from novel intra-
seasonal studies on broad-leaf trees. Plant Cell Environ 27:367–
380
Hemming D, Fritts H, Leavitt SW, Wright W, Long A, Shashkin A
(2001) Modelling tree-ring d13C. Dendrochronologia 19:23–38
Hill SA, Waterhouse JS, Field EM, Switsur VR, Aprees T (1995)
Rapid recycling of triose phosphates in oak stem tissue. Plant
Cell Environ 18:931–936
Hoell W (1985) Seasonal fluctuations of reserve materials in the
trunkwood of spuce (Picea abies (L.) Karst.). J Plant Physiol
117:355–362
Hogberg P, Johannisson C, Hallgren JE (1993) Studies of C-13 in the
foliage reveal interactions between nutrients and water in forest
fertilization experiments. Plant Soil 152:207–214
744 Oecologia (2009) 161:729–745
123
Holmes RL (1983) Computer-assisted quality control in tree-ring
dating and measurement. Tree Ring Bull 44:69–75
Holmes RL (1992) Program COFECHA: Version 3. The University of
Arizona, Tucson
Hubbard RM, Stiller V, Ryan MG, Sperry JS (2001) Stomatal
conductance and photosynthesis vary linearly with plant hydraulic
conductance in ponderosa pine. Plant Cell Environ 24:113–121
Hughes MK, Vaganov EA, Shiyatov SG, Touchan R, Funkhouser G
(1999) Twentieth-century summer warmth in northern Yakutia
in a 600-year context. Holocene 9:603–608
Jones PD, Briffa KR (1992) Global surface air temperature variations
during the twentieth century: part 1, spatial, temporal and
seasonal details. Holocene 2:165–179
Kagawa A, Naito D, Sugimoto A, Maximov TC (2003) Effects of
spatial and temporal variability in soil moisture on widths and
d13C values of eastern Siberian tree rings. J Geophys Res
108:4500
Kagawa A, Sugimoto A, Maximov TC (2006a) (CO2)-C13 pulse-
labelling of photoassimilates reveals carbon allocation within
and between tree rings. Plant Cell Environ 29:1571–1584
Kagawa A, Sugimoto A, Maximov TC (2006b) Seasonal course of
translocation, storage and remobilization of C-13 pulse-labeled
photoassimilate in naturally growing Larix gmelinii saplings.
New Phytol 171:793–804
Keel SG, Siegwolf RTW, Jaggi M, Koerner C (2007) Rapid mixing
between old and new C pools in the canopy of mature forest
trees. Plant Cell Environ 30:963–972
Kirdyanov A, Hughes M, Vaganov E, Schweingruber F, Silkin P
(2003) The importance of early summer temperature and date of
snow melt for tree growth in the Siberian Subarctic. Trees
17:61–69
Knohl A, Schulze ED, Kolle O, Buchmann N (2003) Large carbon
uptake by an unmanaged 250-year-old deciduous forest in
Central Germany. Agric For Meteorol 118:151–167
Lambers H, Chapin FSIII, Pons TL (1998) Plant physiological
ecology. Springer, New York, p 540
Leavitt SW (1993) Environmental information from 13C/12C ratios of
wood. Geophys Monogr 78:325–331
Leikola M (1969) The influence of environmental factors on the
diameter growth of young trees. Acta For Fenn 92:1–44
Linder S (1995) Foliar analysis for detecting and correcting nutrient
imbalances in Norway spruce. Ecol Bull (Copenhagen) 44:178–
190
Livingston NJ, Guy RD, Sun ZJ, Ethier GJ (1999) The effects of
nitrogen stress on the stable carbon isotope composition, produc-
tivity and water use efficiency of white spruce (Picea glauca(Moench) Voss) seedlings. Plant Cell Environ 22:281–289
Lundmark T, Bergh J, Strand M, Koppel A (1998) Seasonal variation
of maximum photochemical efficiency in boreal Norway spruce
stands. Trees 13:63–67
McCarroll D, Loader NJ (2004) Stable isotopes in tree rings. Quat Sci
Rev 23:771–801
McNulty SG, Swank WT (1995) Wood delta-C-13 as a measure of
annual basal area growth and soil-water stress in a pinus-strobus
forest. Ecology 76:1581–1586
Mikola P (1962) Temperature and tree growth near the northern
timberline. In: Kozlowski TT (ed) Tree growth. Ronald, New
York, pp 265–287
Monserud RA, Marshall JD (2001) Time-series analysis of delta C-13
from tree rings I. Time trends and autocorrelation. Tree Physiol
21:1087–1102
Munro MAR, Brown PM, Hughes MK, Garcia EMR (1996) Image
analysis of tracheid dimensions for dendrochronological use. In:
Dean JS, Meko DM, Swetnam TW (eds) Tree rings, environment
and humanity radiocarbon. Birkhauser, Basel, pp 843–853
Oren R, Schulze ED (1989) Nutritional disharmony and forest
decline: a conceptual model. Ecol Stud 77:425–443
Pate J, Arthur D (1998) d13C analysis of phloem sap: novel means of
evaluating seasonal water stress and interpreting carbon isotope
signatures of foliage and trunk wood of Eucalyptus globulus.
Oecologia 117:301–311
Rinn F (1996) Tsap V 3.6 reference manual: computer program for
tree-ring analysis and presentation. RINNTECH, Heidelberg
Scartazza A, Mata K, Matteucci G, Yakir D, Moscatello S, Brugnoli E
(2004) Comparisons of d13C of photosynthetic products and
ecosystem respiratory CO2 and their responses to seasonal
climate variability. Oecologia 140:340–351
Schulze B, Wirth C, Linke P, Brand WA, Kuhlmann I, Horna de
Zimmermann V, Schulze ED (2004) Laser-Ablation-Combus-
tion-GC-IRMS—a new method for online analysis of intra-
annual variation of d13C in tree ring. Tree Physiol 24:1193–1201
Schulze ED, Beck E, Muller-Hohenstein K (2006) Plant ecology.
Springer, Heidelberg, p 846
Schweingruber FH (1988) Tree ring: basics and applications of
dendrochronology. Reidel Publishing, Dordrecht
Schweingruber FH (1996) Tree rings and environment. Dendroecol-
ogy. Paul Haupt, Bern
Schweingruber FH, Boerner A, Schulze ED (2007) Atlas of woody
plant stems. Springer, Heidelberg
Shiyatov SG (1986) Dendrochronology of upper timberline in ural
mountains (in Russian). Nauka, Moscow
Skomarkova MV, Vaganov EA, Mund M, Knohl A, Linke P, Boerner
A, Schulze ED (2006) Inter-annual and seasonal variability of
radial growth, wood density and carbon isotope ratios in tree
rings of beech (Fagus sylvatica) growing in Germany and Italy.
Trees 20(5):571–586
Sudachkova NE, Milyutina IL, Romanova LI, Semenova GP (2004)
The annual dynamics of reserve compounds and hydrolytic
enzyme activity in the tissues of Pinus sylvestris L and Larixsibirica Lebed. Eurasian J For Res 7:1–10
Vaganov EA (1990) The traheidogram method in tree-ring analysis
and its application. In: Cook ER, Kairiuktis LA (eds) Methods of
dendrochronology. Applications in the environmental sciences.
Kluwer, Dordrecht, pp 63–75
Vaganov EA, Shashkin AV, Sviderskaya IV, Vysotskaya LG (1985)
Histometric analysis of woody plant growth (in Russian). Nauka,
Novosibirsk
Vaganov EA, Shiyatov SG, Mazepa VS (1996) Dendroclimatic study
in Ural-Siberian Subarctic (in Russian). Nauka, Novosibirsk
Vaganov EA, Hughes MK, Shashkin EA (2006) Growth dynamics of
conifer tree rings: images of past and future environments.
Springer, Heidelberg
Von Felten S, Hattenschwiler S, Saurer M, Siegwolf R (2007) Carbon
allocation in shoots of alpine treeline conifers in a CO2 enriched
environment. Trees 21:283–294
Walcroft AS, Silvester WB, Whitehead D, Kelliher FM (1997)
Seasonal changes in stable carbon isotope ratios within annual
rings of Pinus radiata reflect environmental regulation of growth
processes. Aust J Plant Physiol 24:57–68
Walter H, Lieth H (1967) Klimadiagramm Weltatlas. Fischer, Jena
Wilson AT, Grinsted WA (1977) 12C/13C in cellulose and lignin as
paleothermometers. Nature 265:133–135
Wimmer RP (2002) Wood anatomical features in tree-rings as
an indicators of environmental changes. Dendrochronologia
20:21–36
Oecologia (2009) 161:729–745 745
123